Real Time Stress Detection in Humans using Smartphone Sensor Data

Stress is a psychological state induced by the body's response to challenging conditions, and it can have a detrimental impact on physical and mental health if experienced for an extended period. Continuous stress in humans leads to various diseases such as diabetes, cardiovascular disease, cancer, hypertension, arthritis, and substance addiction in some instances. Early stress identification is critical for avoiding chronic health issues. Stress identification necessitates continuous monitoring of daily activities to identify the pattern of behaviour in response to various stress-related situations in day-to-day life. The proposal aims to identify the stress in humans using the data collected from a smart watch and a mobile phone through machine learning approaches. Initially, the WESAD dataset is trained and tested for identification of stress using a pool of Machine Learning algorithms such as Random Forest classifier, Logistic regression, Linear Regression, and Support Vector Machine. Subsequently, real-time data such as heart rate collected from a smart watch and mobile pickup time from an Android mobile is used to test the approach's effectiveness. It was found that Random Forest outperforms all the other algorithms, with an accuracy of 99% in identifying stress in individuals.